npct/pct-team

Slides, test code, figures and ideas behind the PCT

pct-team's Issues

Add hilliness variable per route-allocated OD pair

This is data provided by http://www.cyclestreets.net/api/v1/journey/#sampleJSONresult :

 "elevations":"6,6,6,6,6,6,6",

We should add that to the flows.

@usr110 are you up for taking a look at this?

Robin

Rewriting front page text

Here's my first go at proposed changes to the front page text so we can fit in the video... I think we could actually cut it right down - at the moment we provide a little information but the video is better at getting that across.

Welcome to the Propensity to Cycle Tool (PCT)

Created by the Propensity to Cycle Tool team.

Last updated: 2016-07-27.

Welcome to the Propensity to Cycle Tool for England (PCT-England), which provides an evidence base to inform cycling investment. To find out more about the tool, its use and the four scenarios, please watch the video below. More detailed information can be found on PCT blog, the Manual, an academic paper or the about page [include links].

To run the PCT, click on a region in the map to the left. This will open up the PCT for that region. The PCT will then show you cycling levels in that region at area, ‘desire line’ and route network levels, from 2011 Census for commuting and for various scenarios that illustrate where cycling has the greatest potential to grow. This is based on a model that estimates cycling potential based upon trip distances and hilliness.

Currently the PCT only shows commuting trips but additional layers are planned.

Link to PCT blog on main map page

Hi @Robinlovelace @nikolai-b @JDWoodcock - it would be great to get a link to the PCT blog on the main map page, now that I've got a few posts up there including our LTT piece. Ideally I think an additional link after FAQs would be great, but one other option would be to have a 'more info' tab/dropdown that would contain links to 'About', 'FAQs', and the blog.

Start/finish points that are not routable due to OSM data

The following locations are not routable, due to them being in the midst of non-cycleable areas which are technically routable locally by bicycle but which can't be exited from, e.g. motorways or one-way systems.

I recommend these be manually adjusted, as routes from/to these specific locations will fail.

• -0.39483,51.487404 (Bicycles not denied from service station but surrounding connections are all motorways) http://www.openstreetmap.org/?mlat=51.48740&mlon=-0.39484#map=19/51.48740/-0.39484
• -1.574460,54.772446 (Resolves to footpath network that seems not to have any non-gated entry points) http://www.openstreetmap.org/?mlat=54.77245&mlon=-1.57446#map=19/54.77245/-1.57446

I used the script created by @Robinlovelace in 4a96267 and ran this through our batch routing system, which picks up these cases quickly.

This is not at this stage a complete list as I tried the batch on a server that only has about half of the UK loaded. I will edit this opening posting if more are found in future.

(It's on our list to fix these rare cases at engine level, but it's non-trivial because the local routability remains valid. Essentially we need a re-entrant algorithm that can invalidate a nearest point lookup if the other end is on a different network island and then second time around treat the local-only network as if it were bicycle=no even though it the data is bicycle=yes. I have a feeling that Simon has been doing some work on this general area recently but we don't have a stable engine implementation yet.)

List of local files

• ~/Documents/Transport.bib in flow-model.Rmd
• /home/robin/Dropbox/DfT bid/figures/las-gender-pcycle.png in flow-model.Rmd
• /home/robin/Dropbox/DfT bid/Data analysis/Test_DD/TestPlanck_150303/150304_SegmentationCompare_8cat.xlsx in testing-dd-ffs.Rmd and dd-paper.Rmd
• /media/robin/data/data-to-add/msoa-2011-sex-method-ttw-nomis-lc7103ew.csv in load-cents.R
• /media/robin/data/data/UKDA-5340-spss/spss/spss19/ in pswitch-model-NTS.R/blob/master/models/pswitch-model-NTS.R)
• /media/robin/data/data-to-add/public-flow-data-msoa/wu03ew_v2.csv in a few (case-studies/manchester.R, case-studies/leeds-minitest.R, case-studies/leeds-wider.R, and case-studies/manchester.md)

and others. If there are old files in here then let's get rid of them and clean it up.
I think we should aim to get no links to external files included.

Remove beginning and end of lines before generating rnet

Related to npct/pct-shiny#89

How to deal with cua names containing spaces

A few examples:

"North East Lincolnshire" "North Lincolnshire"
[103] "North Somerset" "North Tyneside"

Not sure how best to deal with these in buildmaster.R or load.Rmd

Hilliness basemap is not working

Update readme and documentation so it's easier to reproduce

@nikolai-b, you up for a 1st bash?

Use time not distance and main determinant factor

Add network analysis data

Add % by car variable

This is not a priority issue but would be a useful way to see the potential congestion benefit of cycling uptake in different areas.

Add screenshots of each route viz option

Add 'business' variable to 'fastest' and 'quietest' lines

We should be able to provide an indication of 'business' to routes. This will help estimate the shift to cycling that could happen if an off-road route were to build - it exists here: http://www.cyclestreets.net/api/v1/journey/

 busynance int
    A measure of the cost of busyness of the route, in metres.

Private Data and Missing Files in load.rmd file

@Robinlovelace
As the title says, we can't run the code unless we get access to the private-data folder - mentioned at https://github.com/npct/pct/edit/master/load.Rmd#L129

Also, the shape object, msoasmapshaped_25%.shp, mentioned at:

https://github.com/npct/pct/edit/master/load.Rmd#L62

is not present in the pct-bigdata repo. Would be great if you resolve these issues.

Clean up repo

Add high incline % as predictive variable

Route distances seem to out by factor of 1000

put case studies on pct.bike

@RachelAldred @usr110 @Robinlovelace

as requested by DfT.